instream concentrations of both total nitro- gen (TN) and ortho-P. For this comparison, the SWAT OWTS biozone algorithm—which is designed to simulate nitrogen, phosphorus, bacteria, and biological oxygen demand dis- charges from septic tank e¡uent (Jeong et al., 2011)—was not used and was replaced with the MEANSS estimates. The Prickly Pear watershed in central Mon- tana (Figure 2) was chosen for this project because it has a su§cient number of OWTS (approximately 1,010) for the size of the water- shed (531 km 2 ) to create noticeable impacts to stream water quality. In addition, there is little industrial or agricultural development in this watershed above the USGS streamflow gage near the town of Clancy that could potentially mask the impacts from OWTS. The SWAT model was developed using available information for elevation, land use and land cover, soils, and streamflow. The hydrology was calibrated to daily streamflow values measured near Clancy, Montana, at the USGS Prickly Pear Creek gage (06061500), which was also used as the outlet for the model. The streamflow calibration period was from 1992 through April 2013. Daily mea- sured streamflow was available for 82% of the calibration period. The daily error statistics of relative error, coe§cient of determination, and Nash–Sutcli©e e§ciency coe§cient were -9.0%, 0.76, and 0.76, respectively. All three statistics indicate a good match between mea- sured and simulated streamflow values. The instream water quality calibration data consisted of 20 TN and ortho-P sam- ples collected from 1999 through 2003 by USGS, along with three cold-weather samples (February, March, and April) collected by the Montana Department of Environmental Quality in 2013. Incorporating the steady-state MEANSS loading estimates into the daily-time step SWAT model showed that the lack of sea- sonal variation in MEANSS results created unreasonably large TN and ortho-P values in the winter months during baseflow con- ditions. To provide better seasonal variation for the MEANSS results, it was assumed that OWTS TN and ortho-P contributions to streams varied proportionally with stream- flow. This approach assumes a higher vol- ume of groundwater contribution (and cor- responding OWTS e¡uent contributions) to streams during the spring when ground-
TABLE 3
Validation Results of the Method for Estimating Attenuation of Nutrients From Septic Systems (MEANSS)
Site
Parameter
OWTS Load (or % Reduction) Estimated From Site Information
OWTS Load (or % Reduction) Estimated From MEANSS
% Difference
1 2 3 4
TN TN TN TN TN TN TN
7,008 kg/year 49,551 kg/year
5,312 kg/year 37,767 kg/year
-24 -24
237 kg/year 637 kg/year
289 kg/year 508 kg/year
22
-20
ArcNLET STUMOD
19,023 kg/year
17,956 kg/year
-6
Variable
Variable
*
SWAT SWAT
16.3 kg/day 0.39 kg/day
22.4 kg/day 0.54 kg/day
37 38
Ortho-P
*See Figure 3. Note. ArcNLET = ArcGIS-Based Nitrate Load Estimation Toolkit; ortho-P = ortho-phosphorus; OWTS = onsite wastewater treatment system; STUMOD = Soil Treatment Unit Model; SWAT = Soil and Water Assessment Tool; TN = total nitrogen.
used with the data from Site 2 (Rios et al., 2013). ArcNLET is a GIS-based program that estimates nitrate reduction from OWTS using groundwater velocity rates (calculated from site-specific hydraulic conductivity, hydrau- lic gradient, and porosity) and a user-defined denitrification rate. The same OWTS spatial information used for the MEANSS analysis was also used for ArcNLET. For the ArcNLET analysis, the hydraulic conductivities and hydraulic gra- dient from Miller (1991) were used. The hydraulic conductivity ranged from between 610 and 4,417 m/day, the hydraulic gradi- ent ranged from 0.001 to 0.003 m/m, and a porosity of 25% was estimated using the upper end of the range for sand and gravel aquifers (Driscoll, 1986). Moreover, the denitrification rate suggested in the ArcN- LET documentation (0.008 L/day) was used. Using those parameters, ArcNLET was used to estimate a total nitrate load to the Bitter- root River of 19,067 kg/year. All the MEANSS nitrate scoring categories were used for this comparison. STUMOD MEANSS was compared to a mechanis- tic model, Soil Treatment Unit Model (STUMOD), that calculates nitrogen reduc- tion below the drainfield in the vadose zone
(McCray et al., 2010). To provide comparable results, only the soil type at the drainfield category in MEANSS was used in the com- parison (Table 1). The 12 soil types included in McCray et al. (2010) were compared and the HSG for each of those soil types was esti- mated for comparison purposes. The nitrogen reduction values for STUMOD were estimated from cumulative probability graphs of Monte Carlo simulation results for a deep-water table (McCray et al., 2010). The STUMOD results were based on the following parameters: • hydraulic loading rate: 2 cm/day; •frigid/cryic temperature range (0–8 °C), which is comparable to Montana tempera- ture ranges (Supplemental Figure 1); • reduction estimated at 120-cm soil depth; • standard drainfield e¡uent of 60 mg/L as ammonium-N and 1 mg/L as nitrate-N; and • 50% value on the cumulative frequency distribution (i.e., one half the simulations showed greater reductions and one half the simulations showed lesser reductions). SWAT Model (Prickly Pear Watershed, Montana) The final validation method was a watershed model created using SWAT (Arnold et al., 1993). For this method, OWTS loading val- ues from MEANSS were incorporated into the SWAT simulation and calibrated to observed
13
April 2024 • our9-l o2 9@5ro9me9>-l e-l>4
Powered by FlippingBook